Merge branch 'master' into st/test_Matern_AD

Author: st--

Project.toml

@@ -1,6 +1,6 @@
 name = "KernelFunctions"
 uuid = "ec8451be-7e33-11e9-00cf-bbf324bd1392"
-version = "0.10.35"
+version = "0.10.36"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"

README.md

@@ -9,14 +9,14 @@
 [![DOI](https://zenodo.org/badge/188430419.svg)](https://zenodo.org/badge/latestdoi/188430419)
 
 
-
 ## Kernel functions for machine learning
 
-KernelFunctions.jl provides a flexible framework for defining kernel functions, and an extensive collection of implementations.
-
-The aim is to make the API as model-agnostic as possible while still being user-friendly, and to interoperate well with generic packages for handling parameters like [ParameterHandling.jl](https://github.com/invenia/ParameterHandling.jl/) and FluxML's [Functors.jl](https://github.com/FluxML/Functors.jl/).
-
-Where appropriate, kernels are AD-compatible.
+**KernelFunctions.jl** is a general purpose [kernel](https://en.wikipedia.org/wiki/Positive-definite_kernel) package.
+It provides a flexible framework for creating kernel functions and manipulating them, and an extensive collection of implementations.
+The main goals of this package are:
+- **Flexibility**: operations between kernels should be fluid and easy without breaking, with a user-friendly API.
+- **Plug-and-play**: being model-agnostic; including the kernels before/after other steps should be straightforward. To interoperate well with generic packages for handling parameters like [ParameterHandling.jl](https://github.com/invenia/ParameterHandling.jl/) and FluxML's [Functors.jl](https://github.com/FluxML/Functors.jl/).
+- **Automatic Differentiation compatibility**: all kernel functions which _ought_ to be differentiable using AD packages like [ForwardDiff.jl](https://github.com/JuliaDiff/ForwardDiff.jl) or [Zygote.jl](https://github.com/FluxML/Zygote.jl) _should_ be.
 
 ## Examples
 
@@ -50,9 +50,9 @@ plot(
 
 ## Related Work
 
-Directly inspired by the [MLKernels](https://github.com/trthatcher/MLKernels.jl) package.
+This package replaces the now-defunct [MLKernels.jl](https://github.com/trthatcher/MLKernels.jl). It incorporates lots of excellent existing work from packages such as [GaussianProcesses.jl](https://github.com/STOR-i/GaussianProcesses.jl), and is used in downstream packages such as [AbstractGPs.jl](https://github.com/JuliaGaussianProcesses/AbstractGPs.jl), [ApproximateGPs.jl](https://github.com/JuliaGaussianProcesses/ApproximateGPs.jl), [Stheno.jl](https://github.com/willtebbutt/Stheno.jl), and [AugmentedGaussianProcesses.jl](https://github.com/theogf/AugmentedGaussianProcesses.jl).
 
-See the JuliaGaussianProcesses [Github organisation](https://github.com/JuliaGaussianProcesses) and [website](https://juliagaussianprocesses.github.io/) for more related packages.
+See the JuliaGaussianProcesses [Github organisation](https://github.com/JuliaGaussianProcesses) and [website](https://juliagaussianprocesses.github.io/) for more information.
 
 ## Issues/Contributing
 

docs/src/index.md

@@ -1,10 +1,10 @@
 # KernelFunctions.jl
 
 **KernelFunctions.jl** is a general purpose [kernel](https://en.wikipedia.org/wiki/Positive-definite_kernel) package.
-It aims at providing a flexible framework for creating kernels and manipulating them.
+It provides a flexible framework for creating kernel functions and manipulating them, and an extensive collection of implementations.
 The main goals of this package are:
-- **Flexibility**: operations between kernels should be fluid and easy without breaking.
-- **Plug-and-play**: including the kernels before/after other steps should be straightforward.
+- **Flexibility**: operations between kernels should be fluid and easy without breaking, with a user-friendly API.
+- **Plug-and-play**: being model-agnostic; including the kernels before/after other steps should be straightforward. To interoperate well with generic packages for handling parameters like [ParameterHandling.jl](https://github.com/invenia/ParameterHandling.jl/) and FluxML's [Functors.jl](https://github.com/FluxML/Functors.jl/).
 - **Automatic Differentiation compatibility**: all kernel functions which _ought_ to be differentiable using AD packages like [ForwardDiff.jl](https://github.com/JuliaDiff/ForwardDiff.jl) or [Zygote.jl](https://github.com/FluxML/Zygote.jl) _should_ be.
 
 This package replaces the now-defunct [MLKernels.jl](https://github.com/trthatcher/MLKernels.jl). It incorporates lots of excellent existing work from packages such as [GaussianProcesses.jl](https://github.com/STOR-i/GaussianProcesses.jl), and is used in downstream packages such as [AbstractGPs.jl](https://github.com/JuliaGaussianProcesses/AbstractGPs.jl), [ApproximateGPs.jl](https://github.com/JuliaGaussianProcesses/ApproximateGPs.jl), [Stheno.jl](https://github.com/willtebbutt/Stheno.jl), and [AugmentedGaussianProcesses.jl](https://github.com/theogf/AugmentedGaussianProcesses.jl).

src/test_utils.jl

@@ -94,18 +94,18 @@ function test_interface(
     rng::AbstractRNG, k::Kernel, ::Type{Vector{T}}; kwargs...
 ) where {T<:Real}
     return test_interface(
-        k, randn(rng, T, 1001), randn(rng, T, 1001), randn(rng, T, 1000); kwargs...
+        k, randn(rng, T, 11), randn(rng, T, 11), randn(rng, T, 13); kwargs...
     )
 end
 
 function test_interface(
-    rng::AbstractRNG, k::MOKernel, ::Type{Vector{Tuple{T,Int}}}; dim_out=1, kwargs...
+    rng::AbstractRNG, k::MOKernel, ::Type{Vector{Tuple{T,Int}}}; dim_out=3, kwargs...
 ) where {T<:Real}
     return test_interface(
         k,
-        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:51],
-        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:51],
-        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:50];
+        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:11],
+        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:11],
+        [(randn(rng, T), rand(rng, 1:dim_out)) for i in 1:13];
         kwargs...,
     )
 end
@@ -115,9 +115,9 @@ function test_interface(
 ) where {T<:Real}
     return test_interface(
         k,
-        ColVecs(randn(rng, T, dim_in, 1001)),
-        ColVecs(randn(rng, T, dim_in, 1001)),
-        ColVecs(randn(rng, T, dim_in, 1000));
+        ColVecs(randn(rng, T, dim_in, 11)),
+        ColVecs(randn(rng, T, dim_in, 11)),
+        ColVecs(randn(rng, T, dim_in, 13));
         kwargs...,
     )
 end
@@ -127,9 +127,9 @@ function test_interface(
 ) where {T<:Real}
     return test_interface(
         k,
-        RowVecs(randn(rng, T, 1001, dim_in)),
-        RowVecs(randn(rng, T, 1001, dim_in)),
-        RowVecs(randn(rng, T, 1000, dim_in));
+        RowVecs(randn(rng, T, 11, dim_in)),
+        RowVecs(randn(rng, T, 11, dim_in)),
+        RowVecs(randn(rng, T, 13, dim_in));
         kwargs...,
     )
 end
@@ -139,9 +139,9 @@ function test_interface(
 ) where {T<:Real}
     return test_interface(
         k,
-        [randn(rng, T, dim_in) for _ in 1:1001],
-        [randn(rng, T, dim_in) for _ in 1:1001],
-        [randn(rng, T, dim_in) for _ in 1:1000];
+        [randn(rng, T, dim_in) for _ in 1:11],
+        [randn(rng, T, dim_in) for _ in 1:11],
+        [randn(rng, T, dim_in) for _ in 1:13];
         kwargs...,
     )
 end